As expectations are growing for the Fed to provide further guidance on the exit scenario (for rates and asset purchases), this post provides a short view on what is driving long-term rates and where we may stand in the medium run.
Even after the outbreak of the financial crisis in 2008, long-term interest rates have followed short term interest rates and the economic momentum. However, when the Fed reached the zero bound on rates, the only macro driver of rates that remained was the economic momentum. This feature held well until the implementation of "Operation Twist" (as is clearly visible on the chart below where the relationship between economic news/surprises and 10-year Treasury bonds broke in late 2011).
Finding the drivers
Right now, the US bond market's main features are:
- A weaker link with the economic cycle;
- A disconnect with the short-end (ED1-ED5 vs. 10-year yields). Short rates are being driven by forward guidance, and the long-end, by Fed's asset purchases, among other things;
3. An inverted relationship of the slope of the yield curve with respect to equity returns (explained by the stickiness of the short end); and
4. The link between long yields and the USD has changed. There was never a strong relationship between both assets. The correlation turned negative after QE: positive changes in US yields coming along with negative ones for the dollar. The correlation could return to neutral or even positive territories but for different reasons - yields through a tighter monetary policy, USD through better external accounts.
In the short run though, 3-month changes in yields are still well correlated to 3 months returns of equity markets. In spite of the huge amount of liquidity, there is still a short run arbitrage between stocks and bonds.
Breaking Down Yields
- The Fed likes to break up yields into inflation expectations, real expected short term interest rates, and a risk premium (chart below). The problem is that the term premium encompasses too many factors/parameters, so the use is limited.
2. A glimpse at a chart below shows that TIPS 10-year breakeven is close to 2.6%: inflation has to be at or above this level for the next 10 years for investors to break even, which is quite unlikely in my view. higher yield should not be driven my a much higher inflation premium.
3. Instead of looking at short term real yields I like to compare the public deficit to long term real yields. This reflects the strain on domestic saving due to government borrowing and the associated theoretical and empirical impact on real interest rates. As can be seen below, real rates should be at least 100 bps higher if the Fed did not help financing the deficit
4. Lastly, I like to model US yields using traditional factors (Fed Fund rates, ISM manufacturing as a proxy of economic activity, stock market return as proxy for potential asset classes arbitrages, and consumer survey of expected inflation for the term premium). I always rule out "ad hoc" data such as Fed purchases for the residual of the equation will provide me with an impact of the Fed action on rates.
The chart below shows that 10 year yields in the US are below their fair value and that there is a potential of 100 bps increases ceteris paribus - that is without any further tightening action by the Fed.
Dealing with the exit
Forecasting 10-year interest rates in the US requires a few assumptions:
i. Whether the "new" relationships mentioned above will remain, or if there will be a re-normalization scenario (rates and growth, for instance);
ii. How the Fed will proceed in dealing with the duality of its monetary policy: reducing excess liquidity and "normalizing" Fed Fund rates.
a. As the marginal impact of further purchases declines, the cost of the exit (selling bonds, holding them to maturity) increases;
b. Can the amount of excess reserves hinder a quick rise in the Fed Funds rate if the economy starts to show some strong cyclical resilience?
In my view, the normalization of monetary policy would come along with a return to the "traditional" relationship. Long yields will be more sensitive to the economic cycle, while arbitrage between yields and stocks will remain unscathed.
Implementation of the Exit:
To understand the exit we need to understand the new monetary policy framework. Since 2010, the Fed:
a. Borrows from banks by issuing reserves and investing in treasury bonds and MBS;
b. As the Fed pays interest rates on reserves, the interest rate on reserves (IOR) becomes the effective tool of stabilization policy;
c. With this carry trade, the Fed recorded huge profits. The exit could be costly for the economy and the Fed;
d. If the Fed sells assets when rates have risen it will take a loss. If it sells assets too early, it might trigger much higher yields. How do we deal with this?
I think that the Fed will not place too much attention to the fiscal cost of the exit. The reason for this is that it can create a deferred asset account financed by new reserves. It would translate into a cut in remittances paid to the Treasury but not a recapitalization cost.
My central scenario is that the Fed will start reducing the size of purchases in late 2013 and stop buying bonds in mid-2014. Only then will there be discussion on reducing the size of the balance sheet. There is not only one option (outright sales), as the Fed will probably combine:
i. Outright sales
ii. Reverse repos
iii. Holding to maturity
Many Fed members consider that the level of long term yields is linked to the size of the Fed balance sheet. I believe that flows matter much more. Hence my forecast prices - in a real risk of a step up in US yields once bond purchases end, regardless of what tools are used to reduce excess reserves.
Contrary to general opinion, the Fed does not have to dry up excess reserves before it raises interest rates. The charts below show the theoretical view of monetary policy against the effective conduct of monetary operations.
In addition, as can be seen in the chart below, drawn from here, forecasters systemically underestimate future rates (here short rates) in a tightening cycle. therefore, the theoretical level of rates (model-implied, deficit-implied, growth implied, calendar spread-implied) should quickly be reached once Fed purchases end. This calls for US 10-year yield to reach 3.0% by mid-2014 against 2.6% for the consensus.